Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “VECTOR PROCESSING”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,621 records · Page 90

Comparative assessment of LANDSAT-4 MSS and TM data quality for mapping applications in the southeast

The initial objectives of analyses of the MSS data are two-fold: (1) to evaluate the geodetic accuracy of CCT-P data of the test sites; and (2) to improve the geodetic accuracy by additional processing if the original data either do not meet pre-launch specifications or mapping requirements. The location of 45 ground control points (GCP) digitized from 35 U.S. Geological Survey 1:24,000 scale quadrangles (UTM coordinates) were identified in terms of pixel and scan line values. These 46 points are used to establish UTM position error vector distributions in the scene. As an initial check on the geometric reliability of the MSS data, 28 well-distributed GCPs were input to a program which compares the scaled image distances between all possible point pairs with the corresponding map distances and computes the distance differences; that is, the relative positional errors. The relative errors obtained from initial computations averaged about +/- 200 m. These errors could result from a number of sources, including misidentification of GCP locations, UTM coordinate errors introduced by the map digitizing process or errors resulting from data acquisition and geometric processing.

Welch, R.↗

PIXEL ANALYSIS OF PHOTOSPHERIC SPECTRAL DATA. I. PLASMA DYNAMICS

Recent observations of the photosphere using high spatial and temporal resolution show small dynamic features at or below the current resolving limits. A new pixel dynamics method has been developed to analyze spectral profiles and quantify changes in line displacement, width, asymmetry, and peakedness of photospheric absorption lines. The algorithm evaluates variations of line profile properties in each pixel and determines the statistics of such fluctuations averaged over all pixels in a given region. The method has been used to derive statistical characteristics of pixel fluctuations in observed quiet-Sun regions, an active region with no eruption, and an active region with an ongoing eruption. Using Stokes I images from the Vector Spectromagnetograph (VSM) of the Synoptic Optical Long-term Investigations of the Sun (SOLIS) telescope on 2012 March 13, variations in line width and peakedness of Fe i 6301.5 Å are shown to have a distinct spatial and temporal relationship with an M7.9 X-ray flare in NOAA 11429. This relationship is observed as stationary and contiguous patches of pixels adjacent to a sunspot exhibiting intense flattening in the line profile and line-center displacement as the X-ray flare approaches peak intensity, which is not present in area scans of the non-eruptive active region. The analysis of pixel dynamics allows one to extract quantitative information on differences in plasma dynamics on sub-pixel scales in these photospheric regions. The analysis can be extended to include the Stokes parameters and study signatures of vector components of magnetic fields and coupled plasma properties.

79 ASTRONOMY AND ASTROPHYSICS↗

A support vector machine (SVM) for predicting preferred treatment position in radiotherapy of patients with breast cancer

Purpose: NYU 05-181 protocol compared the CT simulation in both supine and prone positions for 400 patients with breast cancer (200 left-breast and 200 right-breast) to identify which setup is better at sparing heart and lung involvement in the treatment process. The results demonstrated that all right-breast patients benefited from the prone treatment position, while for left-breast patients, 85% were better treated prone and 15% were better treated supine. Using the clinical data collected from this protocol, the authors aimed at developing an automated tool capable of identifying which of the left-breast cancer patients are better treated supine without obtaining a second CT scan in the supine position. Methods: Prone CT scans from 198 of the 200 left-breast cancer patients enrolled in NYU 05-181 protocol were deidentified and exported to a dedicated research planning workstation. Three-dimensional geometric features of the organs at risk and tumor bed were extracted. A two-stage classifier was used to classify patients into the prone class or the supine class. In the first stage, the authors use simple thresholding to divide the patients into two groups based on their in-field heart volume. For patients with in-field heart volume {<=}0.1 cc, the prone position was chosen as the preferred treatment position. Patients with in-field heart volume >0.1 cc will be further classified in the second stage by a weighted support vector machine (SVM). The weight parameters of the SVM were adjusted to maximize the specificity [true-supine/(true-supine+false-prone)] at the cost of lowering but still maintaining reasonable sensitivity [true-prone/(true-prone+false-supine)]. The authors used K-fold cross validations to test the performance of the SVM classifier. A feature selection algorithm was also used to identify features that give the best classification performance. Results: After the first stage, 49 of the 198 left-breast cancer patients were found to have >0.1 cc of in-field heart volume. The three geometric features of heart orientation, distance between heart and tumor, and in-field lung were selected by the feature selection algorithm in the second stage of the two-stage classifier to give the best predefined weighted accuracy. The overall sensitivity and specificity of the proposed method were found to be 90.4% and 99.3%, respectively. Using two-stage classification, the authors reduced the proportion of prone-treated patients that need a second supine CT scan down to 16.3/170 or 9.6%, as compared to 21/170 or 12.4% when the authors use only the first stage (thresholding) for classification. Conclusions: The authors' study showed that a feature-based classifier is feasible for predicting the preferred treatment position, based on features extracted from prone CT scans. The two-stage classifier achieved very high specificity at an acceptable expense of sensitivity.

60 APPLIED LIFE SCIENCES↗

Automatic Classification of Biological Targets in a Tidal Channel Using a Multibeam Sonar

Multibeam sonars are widely used for environmental monitoring of fauna at marine renewable energy sites. However, they can rapidly accrue vast volumes of data, which poses a challenge for data processing. Here, using data from a deployment in a tidal channel with peak currents of 1–2 m s –1 , we demonstrate the data-reduction benefits of real-time automatic classification of targets detected and tracked in multibeam sonar data. First, we evaluate classification capabilities for three machine learning algorithms: random forests, support vector machines, and k-nearest neighbors. For each algorithm, a hill-climbing search optimizes a set of hand-engineered attributes that describe tracked targets. Here, the random forest algorithm is found to be most effective—in postprocessing, discriminating between biological and nonbiological targets with a recall rate of 0.97 and a precision of 0.60. In addition, 89% of biological targets are correctly classified as either seals, diving birds, fish schools, or small targets. Model dependence on the volume of training data is evaluated. Second, a real-time implementation of the model is shown to distinguish between biological targets and nonbiological targets with nearly the same performance as in postprocessing. From this, we make general recommendations for implementing real-time classification of biological targets in multibeam sonar data and the transferability of trained models.

16 TIDAL AND WAVE POWER↗

TESS Data Release Notes:Reprocessing of Sectors 14–19, DR30 & DR33

TESS data release 30 (DR30) provides reprocessed data products of Sector 14 to 19. The updated data products were generated using version 4.0 of the science processing pipeline and conform to the final set of data anomaly flags defined over the last two years of TESS data analysis and pipeline development. Data release 33 (DR33) corresponds to a multisector search for transiting planets in the same reprocessed data. A detailed description of the changes in the data products in DR30 and DR33 is discussed in§2, and a brief list of changes is summarized here: The timestamps for 2 minute cadence and FFI data are more accurate. The differences between reprocessed data and previous data releases are less than 2.0 seconds in all cases. Photometric apertures were increased in size for targets with T mag<11. Three new Data Anomaly Flags were added to mitigate the effects of scattered light:–Cadences with strong scattered light signals or saturation effects that corrupt the calibration data are flagged and removed from analysis (bit 15, value 16384, “Bad Calibration Exclude”).–Scattered light data anomaly flags are customized for each target, and flagged automatically based on the local background level (bit 13, value 4096, ”Scattered light flag”).–Cadences with insufficient targets to derive cotrending basis vectors are flagged and the PDCSAP FLUX light curves are set to NULL at these times (bit 16, value 32768, “Insufficient Targets for Error Correction Exclude”). The planet search of the reprocessed light curves produced a different set of TCEs from the original processed data. Although there is a high degree of overlap between the original and reprocessed data (∼83% of targets produced TCEs in common), new TCEs were produced in DR30 and not every TCE from previous data releases was recovered. The same is true of the multisector search results from DR33 compared to DR28

TESS'↗

Automated Extraction of Flow Features

Computational Fluid Dynamics (CFD) simulations are routinely performed as part of the design process of most fluid handling devices. In order to efficiently and effectively use the results of a CFD simulation, visualization tools are often used. These tools are used in all stages of the CFD simulation including pre-processing, interim-processing, and post-processing, to interpret the results. Each of these stages requires visualization tools that allow one to examine the geometry of the device, as well as the partial or final results of the simulation. An engineer will typically generate a series of contour and vector plots to better understand the physics of how the fluid is interacting with the physical device. Of particular interest are detecting features such as shocks, recirculation zones, and vortices (which will highlight areas of stress and loss). As the demand for CFD analyses continues to increase the need for automated feature extraction capabilities has become vital. In the past, feature extraction and identification were interesting concepts, but not required in understanding the physics of a steady flow field. This is because the results of the more traditional tools like; iso-surface, cuts and streamlines, were more interactive and easily abstracted so they could be represented to the investigator. These tools worked and properly conveyed the collected information at the expense of a great deal of interaction. For unsteady flow-fields, the investigator does not have the luxury of spending time scanning only one "snapshot" of the simulation. Automated assistance is required in pointing out areas of potential interest contained within the flow. This must not require a heavy compute burden (the visualization should not significantly slow down the solution procedure for (co-processing environments). Methods must be developed to abstract the feature of interest and display it in a manner that physically makes sense.

Dorney, Suzanne↗

Automated Extraction of Flow Features

Computational Fluid Dynamics (CFD) simulations are routinely performed as part of the design process of most fluid handling devices. In order to efficiently and effectively use the results of a CFD simulation, visualization tools are often used. These tools are used in all stages of the CFD simulation including pre-processing, interim-processing, and post-processing, to interpret the results. Each of these stages requires visualization tools that allow one to examine the geometry of the device, as well as the partial or final results of the simulation. An engineer will typically generate a series of contour and vector plots to better understand the physics of how the fluid is interacting with the physical device. Of particular interest are detecting features such as shocks, re-circulation zones, and vortices (which will highlight areas of stress and loss). As the demand for CFD analyses continues to increase the need for automated feature extraction capabilities has become vital. In the past, feature extraction and identification were interesting concepts, but not required in understanding the physics of a steady flow field. This is because the results of the more traditional tools like; isc-surface, cuts and streamlines, were more interactive and easily abstracted so they could be represented to the investigator. These tools worked and properly conveyed the collected information at the expense of a great deal of interaction. For unsteady flow-fields, the investigator does not have the luxury of spending time scanning only one "snapshot" of the simulation. Automated assistance is required in pointing out areas of potential interest contained within the flow. This must not require a heavy compute burden (the visualization should not significantly slow down the solution procedure for co-processing environments). Methods must be developed to abstract the feature of interest and display it in a manner that physically makes sense.

Dorney, Suzanne↗

Analysis of events with b-jets and a pair of leptons of the same charge in pp collisions at √s = 8 TeV with the ATLAS detector

An analysis is presented of events containing jets including at least one b -tagged jet, sizeable missing transverse momentum, and at least two leptons including a pair of the same electric charge, with the scalar sum of the jet and lepton transverse momenta being large. A data sample with an integrated luminosity of 20.3 fb –1 of pp collisions at √s = 8 TeV recorded by the ATLAS detector at the Large Hadron Collider is used. Standard Model processes rarely produce these final states, but there are several models of physics beyond the Standard Model that predict an enhanced rate of production of such events; the ones considered here are production of vector-like quarks, enhanced four-top-quark production, pair production of chiral b'-quarks, and production of two positively charged top quarks. Eleven signal regions are defined; subsets of these regions are combined when searching for each class of models. In the three signal regions primarily sensitive to positively charged top quark pair production, the data yield is consistent with the background expectation. There are more data events than expected from background in the set of eight signal regions defined for searching for vector-like quarks and chiral b'-quarks, but the significance of the discrepancy is less than two standard deviations. Furthermore, the discrepancy reaches 2.5 standard deviations in the set of five signal regions defined for searching for four-top-quark production. The results are used to set 95% CL limits on various models.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

An efficient and portable SIMD algorithm for charge/current deposition in Particle-In-Cell codes

In current computer architectures, data movement (from die to network) is by far the most energy consuming part of an algorithm (≈20pJ/word on-die to ≈10,000 pJ/word on the network). To increase memory locality at the hardware level and reduce energy consumption related to data movement, future exascale computers tend to use many-core processors on each compute nodes that will have a reduced clock speed to allow for efficient cooling. To compensate for frequency decrease, machine vendors are making use of long SIMD instruction registers that are able to process multiple data with one arithmetic operator in one clock cycle. SIMD register length is expected to double every four years. As a consequence, Particle-In-Cell (PIC) codes will have to achieve good vectorization to fully take advantage of these upcoming architectures. In this paper, we present a new algorithm that allows for efficient and portable SIMD vectorization of current/charge deposition routines that are, along with the field gathering routines, among the most time consuming parts of the PIC algorithm. Our new algorithm uses a particular data structure that takes into account memory alignment constraints and avoids gather/scat;ter instructions that can significantly affect vectorization performances on current CPUs. The new algorithm was successfully implemented in the 3D skeleton PIC code PICSAR and tested on Haswell Xeon processors (AVX2-256 bits wide data registers). Results show a factor of ×2 to ×2.5 speed-up in double precision for particle shape factor of orders 1–3. The new algorithm can be applied as is on future KNL (Knights Landing) architectures that will include AVX-512 instruction sets with 512 bits register lengths (8 doubles/16 singles). Program summary Program Title: vec_deposition Program Files doi:http://dx.doi.org/10.17632/nh77fv9k8c.1 Licensing provisions: BSD 3-Clause Programming language: Fortran 90 External routines/libraries: OpenMP > 4.0 Nature of problem: Exascale architectures will have many-core processors per node with long vector data registers capable of performing one single instruction on multiple data during one clock cycle. Data register lengths are expected to double every four years and this pushes for new portable solutions for efficiently vectorizing Particle-In-Cell codes on these future many-core architectures. One of the main hotspot routines of the PIC algorithm is the current/charge deposition for which there is no efficient and portable vector algorithm. Solution method: Here we provide an efficient and portable vector algorithm of current/charge deposition routines that uses a new data structure, which significantly reduces gather/scatter operations. Vectorization is controlled using OpenMP 4.0 compiler directives for vectorization which ensures portability across different architectures. Restrictions: Here we do not provide the full PIC algorithm with an executable but only vector routines for current/charge deposition. These scalar/vector routines can be used as library routines in your 3D Particle-In-Cell code. However, to get the best performances out of vector routines you have to satisfy the two following requirements: (1) Your code should implement particle tiling (as explained in the manuscript) to allow for maximized cache reuse and reduce memory accesses that can hinder vector performances. The routines can be used directly on each particle tile. (2) You should compile your code with a Fortran 90 compiler (e.g Intel, gnu or cray) and provide proper alignment flags and compiler alignment directives (more details in README file).

97 MATHEMATICS AND COMPUTING↗

Selective Conversion of CO 2 to Methanol on a In 2 O 3– x –TiO 2 (110) Interface: Importance of Oxide–Oxide Interactions

Methanol is a strategic energy vector for the storage and delivery of energy and is a widely used precursor for the synthesis of many high-value chemicals. The hydrogenation of carbon dioxide (CO 2 ) into methanol is a key process in industrial operations. Here, in this study, we show that an oxide-oxide interface generated by a low loading (0.15 ML) of In 2 O 3-x on a TiO 2 (110) substrate has a high activity and selectivity as a catalyst for the CO 2 + 3H 2 → CH 3 OH + H 2 O process. The properties of the In 2 O 3-x -TiO 2 interface under reaction conditions were investigated using a combination of synchrotron-based ambient pressure X-ray photoelectron spectroscopy (AP-XPS), temperature programmed desorption (TPD), and catalytic testing. The In 2 O 3-x overlayer spread out on top of the titania and was rich in defects and O vacancies that activated CO 2 and H 2 as reactants, without destroying CH 3 O and CH 3 OH as reaction products. The In 2 O 3-x /TiO 2 (110) catalyst is at least one order of magnitude more active than bulk indium oxide while maintaining a very high selectivity (~80%) towards methanol production. Under the rich hydrogen environment of methanol synthesis, the oxide-oxide interactions allowed only a partial reduction of the In cations, preventing the formation of metal alloys as seen in the case of catalysts with metal-indium oxide interfaces. Thus, the dispersion of low loadings of In 2 O 3-x on a stable oxide substrate is a valid and low-cost approach for generating efficient catalysts for CO 2 valorization.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Symmetry breaking and electrical frustration during tip-induced polarization switching in the non-polar cut of lithium niobate single crystals

Polarization switching in ferroelectric materials is governed by a delicate interplay between bulk polarization dynamics and screening processes at surfaces and domain walls. Here we explore the mechanism of tip-induced polarization switching in the non-polar cuts of uniaxial ferroelectrics. In this case, in-plane component of polarization vector switches, allowing for detailed observations of resultant domain morphologies. We observe surprising variability of resultant domain morphologies stemming from fundamental instability of formed charged domain wall and associated electric frustration. In particular, we demonstrate that controlling vertical tip position allows the polarity of the switching to be controlled. This represents very unusual form of symmetry breaking where mechanical motion in vertical direction controls the lateral domain growth. The implication of these studies for ferroelectric devices and domain wall electronics are discussed.

36 MATERIALS SCIENCE↗

Automatic log analysis with NLP for the CMS workflow handling

The central Monte-Carlo production of the CMS experiment utilizes the WLCG infrastructure and manages daily thousands of tasks, each up to thousands of jobs. The distributed computing system is bound to sustain a certain rate of failures of various types, which are currently handled by computing operators a posteriori. Within the context of computing operations, and operation intelligence, we propose a Machine Learning technique to learn from the operators with a view to reduce the operational workload and delays. This work is in continuation of CMS work on operation intelligence to try and reach accurate predictions with Machine Learning. We present an approach to consider the log files of the workflows as regular text to leverage modern techniques from Natural Language Processing (NLP). In general, log files contain a substantial amount of text that is not human language. Therefore, different log parsing approaches are studied in order to map the log files’ words to high dimensional vectors. These vectors are then exploited as feature space to train a model that predicts the action that the operator has to take. This approach has the advantage that the information of the log files is extracted automatically and the format of the logs can be arbitrary. In this work the performance of the log file analysis with NLP is presented and compared to previous approaches.

Layer, Lukas↗

Magnetic flux array for spontaneous magnetic reconnection experiments

Experimental investigation of reconnection in magnetized plasmas relies on accurate characterization of the evolving magnetic fields. In experimental configurations where the plasma dynamics are reproducible, magnetic data can be collected in multiple discharges and combined to provide spatially resolved profiles of the plasma dynamics. However, in experiments on spontaneous magnetic reconnection recently undertaken at the Versatile Toroidal Facility at MIT, the reconnection process is not reproducible and all information on the plasma must be collected in a single discharge. This paper describes a newly developed magnetic flux array which directly measures the toroidal component of the magnetic vector potential, A{sub {phi}}. From the measured A{sub {phi}}, the magnetic field geometry, current density, and reconnection rate are readily obtained, facilitating studies of the three-dimensional dynamics of spontaneous magnetic reconnection. The novel design of the probe array allows for accurate characterization of profiles of A{sub {phi}} at multiple toroidal angles using a relatively small number of signal channels and with minimal disturbance of the plasma.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Supervised learning with word embeddings derived from PubMed captures latent knowledge about protein kinases and cancer

Abstract Inhibiting protein kinases (PKs) that cause cancers has been an important topic in cancer therapy for years. So far, almost 8% of >530 PKs have been targeted by FDA-approved medications, and around 150 protein kinase inhibitors (PKIs) have been tested in clinical trials. We present an approach based on natural language processing and machine learning to investigate the relations between PKs and cancers, predicting PKs whose inhibition would be efficacious to treat a certain cancer. Our approach represents PKs and cancers as semantically meaningful 100-dimensional vectors based on word and concept neighborhoods in PubMed abstracts. We use information about phase I-IV trials in ClinicalTrials.gov to construct a training set for random forest classification. Our results with historical data show that associations between PKs and specific cancers can be predicted years in advance with good accuracy. Our tool can be used to predict the relevance of inhibiting PKs for specific cancers and to support the design of well-focused clinical trials to discover novel PKIs for cancer therapy.

59 BASIC BIOLOGICAL SCIENCES↗

Angular quasi-phase-matching

We propose a generalization of quasi-phase-matching of quadratic nonlinear processes that we call angular quasi-phase-matching (AQPM). It corresponds to a propagation of three collinear interacting electromagnetic waves in a periodically poled nonlinear medium at any angle with respect to the grating vector. AQPM directions are analyzed using groups theory.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Oligoethylene-bridged diferrocene on Ag(110): Monolayer structures and adsorbate-induced faceting

The self-assembly of a ferrocene (Fc) derivative, oligoethylene-bridged diferrocene (diFc), Fc(CH{sub 2}){sub 14}Fc, on Ag(110) surface has been investigated by scanning tunneling microscopy (STM) under ultrahigh vacuum. Three ordered structures, the majority {beta} and the minorities {alpha} and {gamma}, are formed at monolayer coverage. In {alpha} and {gamma}, a unit cell contains one molecule and the molecules are parallel to each other so that the distances between Fc groups and between oligoethylene chains are reduced. A unit cell contains five molecules in {beta}: four of them are parallel to each other but not parallel to the fifth. The interaction between diFc and Ag(110), which is relatively strong in comparison to the intermolecular interaction, is dominant for the assembly of the ordered structures. The adsorption of diFc molecules induces the reorganization of substrate steps, which prefer to follow the directions of the superstructure lattice vectors, i.e. ({+-}1,2) and ({+-}3,2) of the Ag(110) surface. The (12 13-1) facet is formed due to step bunching at regions with high step density. By using in situ STM, the process of step reorganization and faceting has been observed in real time. It is concluded that the Ag adatoms play a key role on the substrate reorganization.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Towards three-body unitarity in D{sup +}{yields}K{sup -}{pi}{sup +}{pi}{sup +}

We assess the importance of final state interactions in D{sup +}{yields}K{sup -}{pi}{sup +}{pi}{sup +}, stressing the consistency between two- and three-body interactions. The basic building block in the calculation is a K{pi} amplitude based on unitarized chiral perturbation theory and with parameters determined by a fit to elastic LASS data. Its analytic extension to the second sheet allows the determination of two poles, associated with the {kappa} and the K{sup *}(1430), and a representation of the amplitude based on them is constructed. The problem of unitarity in the three-body system is formulated in terms of an integral equation, inspired in the Faddeev formalism, which implements a convolution between the weak vertex and the final state hadronic interaction. Three different topologies are considered for the former and, subsequently, the decay amplitude is expressed as a perturbation series. Each term in this series is systematically related to the previous one and a resummation was performed. Remaining effects owing to single and double rescattering processes were then added and results compared to FOCUS data. We found that proper three-body effects are important at threshold and fade away rapidly at higher energies. Our model, based on a vector-weak vertex, can describe qualitative features of the modulus of the decay amplitude and agrees well with its phase in the elastic region.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

A Scalable HPC Insider Threat Monitoring System

Cyber Equations, Inc. is reporting on the results of its DOE SBIR Phase I project for the development of a Scalable HPC Insider Threat Monitoring System. This final report to the Department of Energy for Task DE-FOA-0001618 covers the period from 02/21/2017 to 02/20/2018 and summarizes the larger research accomplishments, which can be divided in theoretical and laboratory experiments. One of the many challenges for the U.S. Department of Energy (DOE) is the support of the high-performance computing (HPC) centers while providing quality information services, strategic vision, leadership, technical expertise, and capabilities necessary to enable the DOE to deliver effective, efficient, economical, and secure HPC services. In order to realize a secure DOE laboratory enterprise, DOE facilities operations must counter a wide range of cyber insider threats, from unsophisticated adversaries up to advanced, nation state sponsored actors. The DOE laboratories also presents an enormous insider attack surface of an ever-growing number of computers, networks, and devices all running a myriad of software. In addition, the DOE laboratories are connected to a variety of external organizations which multiplies the possibility of insider threat. During this SBIR project Cyber Equations developed a strategic combination of supervised and unsupervised machine learning techniques to control scale and classify whether entities, such as hosts, actors, or documents, are showing suspicious insider behavior. We conducted experiments using real cyber data to show that changes in observable behavior (e.g., changes in outbound traffic patterns) are indicative of attacks executing (e.g., insiders exfiltrating data, bots calling home, or an attack spreading). The features and methods we use, have been used in the anomaly detection research and development community for a significant amount of time and robust implementations are openly available. K-Means clustering, Decision Trees, and Support Vector Machine based classifiers are readily available through the R language and open source H20 machine learning framework. The Cyber Equations approach is novel in its use of these classifiers combined with Hadoop based parallel processing technology. The five stages of processing used by CyberInsider improve scalability and reduce the high number of false positives normally associated with network anomaly detection. In summary, the Cyber Equations has successfully accomplished all of its Phase I tasks and demonstrated the feasibility of the Insider Threat Monitoring System.

97 MATHEMATICS AND COMPUTING↗